Issue 41 Memory

By WO Team

Sleep and memory consolidation


While the role of sleep in learning and memory has been widely investigated, much remains to be explained and controversy abounds.

The REM connection

Several studies show that REM sleep (a sleep stage characterized by rapid eye movement) is associated with learning complex motor skills such as playing the piano. This kind of learning involves a type of memory called procedural memory.

One study found that a group of volunteers who were learning to play a rule-based video game improved their game scores after a good night’s sleep. Volunteers who were awakened throughout the night interrupting their REM sleep, did not improve their scores.

Volunteers who played a game that featured totally random moves, in which no learning was involved, did not improve their scores even after REM sleep. The interpretation of these results was that REM sleep consolidated the procedural memory, which is needed to learn.

The slow wave connection

However, not all studies suggest that REM sleep is required.

In one study, volunteers learn to find their way around a virtual city on the computer while their brains were being scanned. The scans showed intense activity in the hippocampus, which is characteristic of spatial memory tasks.

The volunteers then went to sleep while scans were taken. Interestingly, the hippocampus showed activity similar to that seen during the learning period, and this activity appeared during slow wave sleep – not during REM sleep.

The next morning the subject’s ability to navigate the virtual city had improved. The brain activity observed implies consolidation of spatial memory during slow wave sleep.

Sleep learning?

Several years ago, it was proposed that people could learn in their sleep by using audiotapes. This possibility has been conclusively set to rest.

Researchers used EEGs to ensure that audio information was given only while subjects were actually asleep, and they found absolutely no evidence of learning the next morning. No amount of repetition made any difference. This is not surprising as the brain shuts itself off from outside stimuli during sleep.

Sleep to remember

How does sleep enhance memory?

Some understanding of this question comes from experiments done by Matthew Walker, a neuroscientist at Harvard Medical School. He studied 100 subjects with two sets of finger tapping exercises.

The learning of the second tapping sequence did not erase the memory of the first one. The subjects performed the first exercise well if they performed it immediately after the training session for the second exercise.

However, only the performance of the second sequence was enhanced after sleeping, not the first. The memory of the second sequence has interfered with the memory of the first.

In contrast, if the second sequence was learned six hours after the first sequence, interference did not occur and both sequences were enhanced by sleep.

This finding demonstrated that the memory of the first exercise consolidated within six hours. However, the memory has to be processed still further if it is to be subsequently enhanced by sleep.

Walker suggests that memory is reactivated into a labile state during sleep that allows for reprocessing (reconsolidation). Probably not all memories undergo this process of reconsolidation but the study shows the dynamic nature of memories.

Critics of the sleep-memory connection point out that there have been people with brain injuries who, although they can get little, if any, REM sleep, are able to carry out activities that require procedural memory.

While there is no simple relationship between REM and memory, REM is probably involved in some types of memory formation and not in others. Although REM sleep may assist memory formation, it isn’t the only way new memories can integrate into the neural memory banks.


Nader, K: Memory traces unbound. Trends Neurosci. 2003, 26: 65-72.

Siegel, JM: The REM sleep-memory consolidation hypothesis. Science, 2001, 294: 1058-1063. Lawton, G: To sleep, perchance to dream. New Scientist, 2003, 28 June, pp. 28-35.